I've configured an agent that retrieves data from text-based knowledge sources, and it does generate an answer. However, once it finishes, it removes the entire reply and instead displays the phrase: "I'm sorry, I'm not sure how to help with that. Can you try rephrasing?"
I noticed that initially the agent triggers the Conversational Boosting topic, but when it finishes, it automatically triggers the Fallback topic. As a result, the initially generated answer is removed. This issue also occurs when the Fallback topic is switched off. The only difference in that case is that the fallback phrase"I'm sorry, I'm not sure how to help with that. Can you try rephrasing?"does not appear.
The only potential cause I can think of is that I instructed the agent to extract data points in a specific format once it receives a particular trigger phrase. That's why I tried to configure a new topic that is triggered by the instructed trigger phrase, but it didn’t solve the problem.
Question: How can I resolve this issue?

Report
All responses (
Answers (